[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Public WebGL] WebGL / Collada demo

On Thu, Mar 10, 2011 at 9:18 AM, Erik Möller <emoller@opera.com> wrote:

Sorry about the cross posting, but the collada list seems to be very low traffic and I figured there would be people on the WebGL list sitting on relevant info.

I'm working on a WebGL / Collada demo and article that will be published on dev.opera.com this spring. It's working pretty well, but there are a few outstanding issues I need to address.

One of the models used is a fairly complex rigged, skinned and normal-mapped character. I'm exporting the model using Maya and the latest FBX plug-in (2011.3.1). Now, is there something I've missed or is there no support in Collada for normal maps? If so what are the plans for adding it in?

Hi Erik,
  We've had much better luck with the OpenCollada (http://opencollada.org/) Maya exporter. That's the one we use for Google Body.  Admittedly we primarily use it to export mesh and material info (normal maps are definitely exported) so I haven't looked at joint hierarchies and animations.  I'm also sure that it will be way easier to report and fix bugs on these open source exporters than it is for the built-in one.


I've also been trying to find a decent Collada viewer to verify that the exported data is actually correct, but haven't been very successful. The ones I've downloaded and tried has only worked for very simple scenes. The best one I've tried to date is the default "Preview"-application shipping with OS X but it too fails on my character model. Any tips on good viewers to try that can deal with joints and multiple skins?

I'm seeing a pretty strange result when exporting the skeleton. A few of the joints are oriented all wrong. Since one of them was very obvious (off by 180 degrees) I managed to work out that by changing the sign of the angle on rotations labeled with "post-rotation" and also changing the rotation order for "post-rotations" it would be correct and lo and behold all the other joints popped into the right place.

Now, since I haven't found a viewer that can show this model I haven't been able to identify if it's a mistake in my code or simply wrong data getting exported so again, pointers on decent functional Collada viewers are greatly appreciated or if you by chance have some great insight in how these "post-rotations" are calculated and used and would like to share it with me that's even better.

I dumped the dae containing just the skeleton here in case someone wants to check if their neat viewer can get it right. http://people.opera.com/~emoller/anim_walk_GDC.dae

Another issue I've come across is that when Maya specifies BEZIER for animation data you get passed two control points and two tangents as opposed to four control points like the spec says. "HERMITE is equivalent to BEZIER, but instead of providing the control points C0 and C1, tangents T0 and T1 are provided." I've never gotten HERMITE back from Maya so I'm not sure if the two types are flipped around or if it's always exporting two control points and two tangents.

Erik Möller
Core Developer
Opera Software
You are currently subscribed to public_webgl@khronos.org.
To unsubscribe, send an email to majordomo@khronos.org with
the following command in the body of your email:
unsubscribe public_webgl